Article
Environmental Sciences
Brindusa Cristina Budei, Benoit St-Onge, Richard A. A. Fournier, Daniel Kneeshaw
Summary: Identifying tree species using multispectral lidar can improve forest management decision-making, but the influence of scan angle on classification accuracy needs to be evaluated. This study found that the correlation between feature values and scan angle was poor, with minimal impact on species classification accuracy.
Review
Environmental Sciences
Maja Michalowska, Jacek Rapinski
Summary: Remote sensing techniques, especially Light Detection and Ranging (LiDAR), have greatly improved large-scale forest inventory by providing three-dimensional point cloud data for object extraction and classification. Various LiDAR-derived metrics, combined with classification algorithms, contribute to high accuracy in tree species discrimination. Full-waveform data extraction and the use of random forest or support vector machine classifiers have shown to be most effective in increasing species discrimination performance.
Article
Chemistry, Analytical
Hongying Zhang, Jinxin He, Shengbo Chen, Ye Zhan, Yanyan Bai, Yujia Qin
Summary: This paper studied the importance of selecting training samples in remote sensing image classification and compared the effectiveness of grouping selection, entropy-based selection, and direct selection. The experimental results showed that the grouping selection method achieved higher classification accuracy using fewer samples and outperformed the direct selection method. Within a certain range, increasing the number of samples can improve the accuracy of image classification.
Article
Engineering, Electrical & Electronic
Svetlana Illarionova, Alexey Trekin, Vladimir Ignatiev, Ivan Oseledets
Summary: By utilizing multispectral satellite imagery and neural networks, the forest classification problem was addressed as an image segmentation task, represented as a hierarchical set of binary classification tasks, to achieve better results.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Environmental Sciences
Kumar Mainali, Michael Evans, David Saavedra, Emily Mills, Becca Madsen, Susan Minnemeyer
Summary: Accurate and up-to-date wetland maps are crucial for landscape scale wetland conservation. This study successfully developed a deep learning model to automatically map wetlands at high resolution using free data and efficient deep learning models that do not require manual feature engineering. This approach can be a game changer in informing restoration and development decisions for wetlands due to the dynamic nature and important ecosystem services they provide.
SCIENCE OF THE TOTAL ENVIRONMENT
(2023)
Article
Forestry
Martin Slavik, Karel Kuzelka, Roman Modlinger, Peter Surovy
Summary: This study proposes a method of tree species classification using individual tree metrics derived from three-dimensional point cloud data obtained by unmanned aerial vehicle laser scanning. The metrics of 1045 trees were evaluated using a generalized linear model and random forest techniques, leading to automated assignment of individual trees into either coniferous or broadleaf groups. The inclusion of a spatial aggregation index called the Clark-Evans index significantly improved classification accuracy, with overall accuracies of 94.8% and 95.1% achieved using the generalized linear model and random forest approaches, respectively.
Article
Forestry
Peng Sun, Xuguang Yuan, Dan Li
Summary: Tree species surveys are crucial in forest resource management and can provide references for forest protection policymakers. Airborne LiDAR technology, especially using the Transformer algorithm, shows promise for accurate and efficient tree species classification. This research provides a theoretical basis and technical support for future research in the field of forest resource supervision based on UAV remote sensing.
Article
Engineering, Electrical & Electronic
Manish Pratap Singh, V Gayathri, Debasis Chaudhuri
Summary: This article discusses the classification of remote sensing images and proposes a preprocessing technique based on local statistics and a postprocessing technique using a simple filter to improve classification accuracy. The results of the proposed method are compared with other classifiers.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Ally Kruper, Robert J. McGaughey, Sarah Crumrine, Bernard T. Bormann, Keven Bennett, Courtney R. Bobsin
Summary: This study combined LiDAR data with field data to improve the accuracy of tree coordinates and differentiate between specific tree species.
Article
Geochemistry & Geophysics
Jun Zhang, Jiao Liu, Bin Pan, Zongqing Chen, Xia Xu, Zhenwei Shi
Summary: The domain adaptation algorithm for remote sensing image scene classification aims to overcome distribution differences between training and test data, but may face challenges with the appearance of new categories in the target domain. Additionally, focusing solely on transferability can reduce classification performance due to high interclass similarity in remote sensing images.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Plant Sciences
Michael M. Bahe, Ryan L. Murphy, Matthew B. Russell, Joseph F. Knight, Gary R. Johnson
Summary: The efficacy of a miniature multispectral, single-sensor camera for detecting stress in deciduous juvenile tree foliage in a controlled environment was validated in this study. Results showed that NDVI values were reliable indicators of tree stress within species groups, but not across species groups.
URBAN FORESTRY & URBAN GREENING
(2021)
Article
Engineering, Multidisciplinary
Liu TianZhu, Gu YanFeng, Jia XiuPing
Summary: Fine classification of large-scale scenes is important in optical remote sensing applications. Multispectral images (MSIs) and hyperspectral images (HSIs) have complementary characteristics. Collaborative classification of multispectral-hyperspectral remote sensing images has become a hot topic. This paper proposes a class-guided coupled dictionary learning method, which shows better classification performance.
SCIENCE CHINA-TECHNOLOGICAL SCIENCES
(2022)
Article
Environmental Sciences
Xiaoyan Zhang, Linhui Li, Donglin Di, Jian Wang, Guangsheng Chen, Weipeng Jing, Mahmoud Emam
Summary: In this paper, we propose an improved squeeze and excitation residual network (SERNet) for the semantic segmentation of high-resolution remote sensing images. The SERNet integrates several squeeze and excitation residual modules (SERMs) and a refine attention module (RAM), which effectively addresses the challenges posed by complex distribution of ground objects and unclear boundaries. Experimental results demonstrate the superior performance of SERNet on the ISPRS datasets.
Article
Geochemistry & Geophysics
Hong-Kyu Shin, Kwang-Hyun Uhm, Seung-Won Jung, Sung-Jea Ko
Summary: Scene classification is an important task in remote sensing, and multispectral (MS) images are crucial for better classification. However, they are not always available due to cost and complexity. To address this, a new MS-to-RGB knowledge distillation (MS2RGB-KD) framework is proposed in this letter. It transfers MS knowledge from a teacher model to a student model, improving scene classification performance using only RGB images.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Environmental Sciences
Wenju Zhao, Chun Zhou, Changquan Zhou, Hong Ma, Zhijun Wang
Summary: Soil salinization severely restricts the development of global industry and agriculture, and affects human beings. This study develops and optimizes an inversion monitoring model for monitoring soil salt content using unmanned aerial vehicle (UAV) multispectral remote sensing data and provides technical support for rapid monitoring and inversion of soil salinization in irrigation areas.
Article
Geochemistry & Geophysics
Petri Varvia, Timo Lahivaara, Matti Maltamo, Petteri Packalen, Timo Tokola, Aku Seppanen
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2017)
Article
Forestry
Blanca Sanz, Jukka Malinen, Vesa Leppanen, Ruben Valbuena, Tuomo Kauranne, Timo Tokola
CANADIAN JOURNAL OF FOREST RESEARCH
(2018)
Article
Forestry
Qing Xu, Bo Li, Matti Maltamo, Timo Tokola, Zhengyang Hou
FOREST ECOLOGY AND MANAGEMENT
(2019)
Article
Environmental Sciences
U. Mattila, T. Tokola
JOURNAL OF ENVIRONMENTAL MANAGEMENT
(2019)
Article
Forestry
Cheikh Mohamedou, Lauri Korhonen, Kalle Eerikainen, Timo Tokola
Article
Forestry
Katalin Waga, Piotr Tompalski, Nicholas C. Coops, Joanne C. White, Michael A. Wulder, Jukka Malinen, Timo Tokola
Article
Environmental Sciences
Zhengyang Hou, Lauri Mehtatalo, Ronald E. McRoberts, Goran Stahl, Timo Tokola, Parvez Rana, Jouni Siipilehto, Qing Xu
REMOTE SENSING OF ENVIRONMENT
(2019)
Article
Forestry
Blanca Sanz, Jukka Malinen, Jussi Heiskanen, Timo Tokola
Article
Forestry
Ville Karjalainen, Timo Tokola, Jukka Malinen
Summary: This study examines the possibility of predicting the stoniness of topsoil using geophysical data and soil type information. The results show the potential of using gamma-ray and soil type data for estimating topsoil stoniness.
CANADIAN JOURNAL OF FOREST RESEARCH
(2022)
Article
Forestry
Qing Xu, Goran Stahl, Ronald E. McRoberts, Bo Li, Timo Tokola, Zhengyang Hou
Summary: Reliable statistical inference is crucial for forest ecology and management, with systematic adaptive cluster sampling (SACS) being an unbiased and efficient method for inventorying spatially clustered populations. However, challenges such as oversampling and uncertainty in sample formation still exist, leading to the development of a generalized SACS (GSACS) which outperforms systematic sampling (SS) in inventorying clustered populations and making domain-specific estimates.
FOREST ECOLOGY AND MANAGEMENT
(2021)
Article
Forestry
Blanca Sanz, Jukka Malinen, Sanna Sirparanta, Jussi Peuhkurinen, Vesa Leppanen, Timo Melkas, Kirsi Riekki, Tuomo Kauranne, Mikko Vastaranta, Timo Tokola
Summary: The presented methodology aims to improve efficiency in timber markets by assessing the value of harvestable timber stands and timber assortments through alternate approaches using various data sources. Evaluation of timber value and assortment distributions through different bucking scenarios can help in identifying valuable stands and optimizing logging recoveries.
Review
Energy & Fuels
Olli-Jussi Korpinen, Mika Aalto, K. C. Raghu, Timo Tokola, Tapio Ranta
Summary: In this paper, we reviewed 94 publications that focused on analyzing and optimizing energy biomass supply chains using spatial data. The study found that there has been an increase in the use of geographical information systems in this field, along with a diversity of methods, objectives, and data sources. However, case studies with spatial data from multiple countries were scarce in the reviewed papers. The paper also calls for the development of a standard way of reporting geographical contents to improve the comprehension and reproducibility of research in this field.
Article
Forestry
Katalin Waga, Jukka Malinen, Timo Tokola
Summary: Two different pulse density airborne laser scanning datasets were used to develop a quality assessment methodology for determining forest road quality, with the use of a reference DEM. The high pulse density dataset provided better classification results than the low pulse density dataset, and the use of a reference DEM increased the precision of road quality classification. The study compared four interpolation techniques and found that spline interpolation provided the best classification results, showing the potential for identifying poor quality road sections for maintenance purposes.
Article
Forestry
Katalin Waga, Jukka Malinen, Timo Tokola